AfshinMA/Traffic_Sign_Detection-Streamlit_App

1# Import required libraries
2import os
3import keras
4import numpy as np
5import pandas as pd
6import streamlit as st
7from PIL import Image
8
9# Function to safely load the models
10def load_model_safely(path: str):
11    if not os.path.isfile(path) or not path.endswith('.keras'):
12        raise FileNotFoundError(f"The file '{path}' does not exist or is not a .keras file.")
13    return keras.saving.load_model(path)
14
15# Retrieve the current directory and specify model paths
16current_dir = os.getcwd()  # Ensure correct initial directory
17model_paths = {
18    'CNN': os.path.join(current_dir, 'models', 'cnn_model.keras'),
19    'VGG19': os.path.join(current_dir, 'models', 'vgg19_model.keras'),
20    'ResNet50': os.path.join(current_dir, 'models', 'resnet50_model.keras'),
21}
22
23# Load models and handle potential exceptions
24models = {}
25for name, path in model_paths.items():
26    try:
27        models[name] = load_model_safely(path)
28    except Exception as e:
29        st.error(f"Error loading model {name} from {path}: {str(e)}")
30
31# Define the class labels
32classes = { 0:'Speed limit (20km/h)', 1:'Speed limit (30km/h)', 2:'Speed limit (50km/h)', 
33            3:'Speed limit (60km/h)', 4:'Speed limit (70km/h)', 5:'Speed limit (80km/h)', 
34            6:'End of speed limit (80km/h)', 7:'Speed limit (100km/h)', 8:'Speed limit (120km/h)', 
35            9:'No passing', 10:'No passing veh over 3.5 tons', 11:'Right-of-way at intersection', 
36            12:'Priority road', 13:'Yield', 14:'Stop', 15:'No vehicles', 
37            16:'Veh > 3.5 tons prohibited', 17:'No entry', 18:'General caution', 
38            19:'Dangerous curve left', 20:'Dangerous curve right', 21:'Double curve', 
39            22:'Bumpy road', 23:'Slippery road', 24:'Road narrows on the right', 
40            25:'Road work', 26:'Traffic signals', 27:'Pedestrians', 28:'Children crossing', 
41            29:'Bicycles crossing', 30:'Beware of ice/snow', 31:'Wild animals crossing', 
42            32:'End speed + passing limits', 33:'Turn right ahead', 34:'Turn left ahead', 
43            35:'Ahead only', 36:'Go straight or right', 37:'Go straight or left', 
44            38:'Keep right', 39:'Keep left', 40:'Roundabout mandatory', 
45            41:'End of no passing', 42:'End no passing veh > 3.5 tons' }
46
47# Function to preprocess the image and predict the class
48def preprocess_and_predict(image: Image.Image, size=(50, 50)) -> pd.DataFrame:
49    img_resized = image.resize(size)
50    img_array = np.array(img_resized).astype(np.float32) / 255.0
51    img_array = np.expand_dims(img_array, axis=0)  # Shape (1, 50, 50, 3)
52
53    predictions = []
54    for name, model in models.items():
55        predicted_class_index = np.argmax(model.predict(img_array), axis=-1)[0]
56        predictions.append({'Model': name, 'Predicted Label': classes[predicted_class_index]})
57
58    return pd.DataFrame(predictions)
59
60# Import Example images
61images_dir = os.path.join(current_dir, 'images')
62
63if os.path.exists(images_dir):
64    # Create a list of images and their corresponding classes
65    image_list = [img for img in os.listdir(images_dir) if img.lower().endswith('.png')]
66    image_dict = {classes[int(img.split('.')[0])] : os.path.join(images_dir, img) for img in image_list}
67else:
68    st.error(f"The images directory does not exist: {images_dir}")
69
70# Streamlit UI setup
71st.set_page_config(page_title="Traffic Sign Detection App", page_icon="🚦", layout="wide")
72st.title("🚦 Traffic Sign Recognition using CNN, VGG19, ResNet50")
73st.markdown("Upload a traffic sign image or choose an example from below to get the recognition result.")
74st.markdown("---")
75
76# Sidebar for image upload and selection
77st.sidebar.header("Input Options")
78uploaded_file = st.sidebar.file_uploader("Upload an image (JPG, JPEG, PNG)", type=["jpg", "jpeg", "png"])
79
80# Select an example image
81selected_example = st.sidebar.selectbox("Or select an example image:", list(image_dict.keys()))
82if selected_example:
83    example_image_path = image_dict[selected_example]
84
85# Initialize a variable to hold the image for prediction
86image_to_predict = None
87
88# Check if user uploaded an image or selected an example image
89if uploaded_file is not None:
90    image_to_predict = Image.open(uploaded_file)
91    st.image(image_to_predict.resize((256, 256)), caption='Uploaded Image', use_container_width=False, output_format="auto")
92elif selected_example:
93    image_to_predict = Image.open(example_image_path)
94    st.image(image_to_predict.resize((256, 256)), caption='Example Image', use_container_width=False, output_format="auto")
95
96# Add a predict button
97if st.sidebar.button("🚀 Predict", key="predict_button") and image_to_predict is not None:
98    # Run prediction
99    st.write("Predicting ...")
100    results = preprocess_and_predict(image_to_predict)
101    
102    # Display results
103    st.write("### Prediction Results")
104
105    # Style the output dataframe
106    st.dataframe(results)
107
108# Add some custom CSS for better styling
109st.markdown("""
110<style>
111    .stButton > button:hover {
112        background-color: #0052cc; /* Darker blue on hover */
113    }
114    .stDataframe {
115        border: 1px solid #ddd;     /* Light border for clarity */
116        border-radius: 10px;        /* Rounded corners for the dataframe */
117    }
118    .stImage {
119        border: 2px solid #0066ff; /* Border for images */
120        border-radius: 10px;       /* Rounded corners */
121        box-shadow: 0 0 8px rgba(0, 0, 0, 0.2); /* Subtle shadow */
122    }
123</style>
124""", unsafe_allow_html=True)